Show Menu
Cheatography
  1. Home >
  2. Education >

R Cheat Sheet V2: Electric Boogaloo Cheat Sheet (DRAFT) by

Version 2 of the R Cheat Sheet for your Final

This is a draft cheat sheet. It is a work in progress and is not finished yet.

Operators

=
Assigns a value to an object
<-
x > y
x greater than y
x < y
x is less than y
x >= y
x greater than or equal to y
x <= y
x is less than or equal to y
!= x
not equal to x
!x
not x
x | y
x OR y
x & y
x AND y

Basic R Functions

Access a function's help file
help(​­fun­​ction name)
Load a csv file
read.c­sv(­"­sna­ils.cs­v", header = TRUE, row.names = NULL)
Install a library
insta​­ll.p­​a­c​k­ag​­es​­(​"​­library name")
Load an installed library
libra​­ry​­(​l­​ibrary name)
Resize images in Jupyter and Google Collab
optio​­ns​­(​r­​ep­​r.p­​l­o​t.width = x, repr.p­​lo­​t.h­​eight = y)
Return the amount of values in x
lengt​h(x)
Return the number of rows in a dataframe
nrow(df)
Return the absolute value(s) in x
abs(x)
Return the sum of all the values in x
sum(x)
Return the square­​-root of the value(s) in x
sqrt(x)
Return the mean of the values in x with optional arguments for trimming and removing NAs
mean(x, tr = 0, na.rm = FALSE)
Return the median of the values in x with optional arguments removing NAs
median(x, na.rm = FALSE)
Return the sample standard deviation of values in x with optional argument for removing NAs
sd(x, na.rm = FALSE)
Return the sample variance of values in x with optional argument for removing NAs
var(x, na.rm = FALSE)
Return the quartiles for x with optional argument for removing NAs
quant​­ile(x, na.rm = FALSE)
Sort the values of x into ascending order
sort(x)
Compute the median absolute deviation of x with optional argument to remove NAs
mad(x, na.rm = FALSE)
Find NA values in x (returns TRUE/F­​ALSE)
is.na(x)
Paste things together into a single string
paste(x, y, z, sep = "­")
Create a table of counts
Examples: 
table(x)

table(x, y)

Data Frames

Create a new data frame
Column_1 = c("A­​", "­​B", "­C") 
Column_2 = c(21, 22, NA)
new_df = data.f­​ra­m​e­(C​­olu­​mn_1, Column_2)
Add a column
new_d​­f$C­​ol­umn_3 = c(51, 52, 53)
Select a specific value (e.g., 52 = row 2, column 3)
new_df[2, 3]
Select a series of values (e.g., all of row 2)
new_df[2, c(1,2,3)]

or
new_df[2, ]
Select an entire column (e.g., column 2)
new_d​­f$C­​ol­u​mn_2

or
new_df[ , 2]
Isolate column values that are not NAs
new_d​­f$C­​ol­u​m­n_​­2[!­​is.n​­a(​­new­​_d­f​$­Co​­lum­​n_2)]

Subset Function

Used to select specific observ­​ations from a dataframe according to a rule you specify.
subse​­t(d­​at­a​f­rame, subset rule, select = ("co­​lumns to keep"))
Example:
outliers = subset­​(h­e​i­gh​­tData, Father < 60.1 | Father > 75.3, select = c("F­​at­h​e­r"))

Library Functions

libr​a­ry​­(car)
Levene's Test
levene­Tes­t(d­ata­_fr­ame­$Re­sponse, data_f­ram­e$P­red­ictor, center = median)
Bootst­rapping a Regression Model
x = Boot(m­odel, R = 2000) 
 hist(x) 
 confint(x) 
 summary(x)
Type III Sum of Squares ANOVA
Anova(­model, type = "­III­")
libr​a­ry​­(ef­​fs­i​ze)
Cohen's d and Hedges g
cohen.d­(y~x, data, hedges.co­​rr­e​ction = FALSE)
libr​a­ry​­(pl​yr)
Aggregate data frames
new_df = ddply(­​da­t​a­frame, c("P­red­ictor1, Predic­tor­2"), summarise, 
   n = length­​(S­c​o­re​­_Co­​lumn), 
   Means = mean(S­cor­e_C­olumn) )
librar­y(p­olycor)
Biserial Correl­ation
polyse­rial(y, x)
librar­y(pwr)
Sample Size for a Two-Sample T-test
pwr.t.t­est(d, sig.level, power, type = c("t­wo.s­ample, "­pai­red­"))
Sample Size for a One-Way ANOVA
pwr.an­ova.te­st(k, f, sig.level, power)
libr​a­ry​­(rc­​om­p​a­ni​on)
Calculates lambda for Tukey's ladder of powers
trans​­for­​mT­u​k­ey(x, plotit = FALSE, return­​Lambda = TRUE)
libr​a­ry​­(WR​S2)
Winsorized variance of x
winvar(x, tr = .2)
Yuen's two sample t-test for trimmed indepe­ndent means
yuen(y ~ x, tr = .2)
One-Way Robust Indepe­ndent ANOVA with bootst­rap­ping: F-tests
t1wayb­t(R­esponse ~ Predictor, data = data, tr = 0.2, nboot = 2000)
One-Way Robust Indepe­ndent ANOVA with bootst­rap­ping: Post Hocs
mcppb2­0(R­esp­onse~ Predictor, data = data, tr = 0.2, nboot = 2000)
Two-Way Robust Indepe­ndent ANOVA: F-tests
t2way(­Res­ponse ~ Predictor A+ Predictor B + Predictor A : Predictor B, data = depress, tr = 0.2)
Two-Way Robust Indepe­ndent ANOVA: Post-Hocs
x = mcp2at­m(R­esponse ~ Predictor A+ Predictor B + Predictor A : Predictor B, data = depress, tr = 0.2) 
 x$cont­rasts 
 x

Distri­bution Functions

Return the the corres­​po­nding quantile for a given probab­​ility
Normal Distri­bution
qnorm​­(pr­​ob­a​b­ility, mean, sd)
T Distri­bution
qt(pr​­oba­​bi­lity, df, lower.t­ail)
F Distri­bution
qf(pro­bab­ility, df1, df2, lower.t­ail)
Chi-Square Distri­bution
qchisq­(pr­oba­bility, df, lower.t­ail)
Return the the corres­​po­nding probab­​ility for a given quantile.
Normal Distri­bution
pnorm​­(qu­antile, mean, sd)
T Distri­bution
pt(qua­ntile, df, lower.t­ail)
F Distri­bution
pf(qua­ntile, df1, df2, lower.t­ail)
Chi-Square Distri­bution
pchisq­(qu­antile, df, lower.t­ail)

Regression and ANOVA Functions

Factoring a Predictor
 
data_f­ram­e$P­red­ictor = factor­(da­ta_­fra­me$­Pre­dictor)
Viewing levels of a factor
 
levels­(da­ta_­fra­me$­Pre­dictor)
Linear Model
 
model = lm(Res­ponse ~ Predictor1 + Predic­tor2, data = data)
Summary output of a linear model
 
summar­y(m­odel)
Linear Model Confidence Intervals
 
confin­t(m­odel)
F-test Model Compar­isons
 
anova(­model1, model2, model3, etc...)
Anova main effects
 
summar­y(a­ov(­model))
Dummy Coding with 1s and 0s
 
ifelse­(da­ta_­fra­me$­Pre­dictor == "­X", 1, 0)
Contrasts
 
cont1 = c(1, 1, -2) 
 cont2 = c(1, -1, 0) 
 contra­sts­(da­ta_­fra­me$­Pre­dictor) = cbind(­cont1, cont2)
Polynomial Contrasts
 
contra­sts­(da­ta_­fra­me$­Pre­dictor) = contr.p­ol­y(l­eve­ls(­dat­a_f­ram­e$P­red­ictor))
Post Hoc Tests ("bo­nfe­rro­ni", "­hol­m", "­BH")
 
pairwi­se.t.t­est­(da­ta_­fra­me$­Res­ponse, data_f­ram­e$P­red­ictor, p.adju­st.m­ethod = c("h­olm­"))
Tukey HSD
 
TukeyH­SD(­aov­(mo­del), "­Pre­dic­tor­")
Note: the 
lm()
function stores many useful things as attrib­utes:
model$­res­iduals

model$­coe­ffi­cients

Common Statis­tical Tests and Calcul­ations

T-test
t.tes​­t(y~x, altern­​ative = c("t­wo.s­id­ed"), mu = 0, var.equal = FALSE, conf.level = 0.95)
Correl­ation
cor(x, y)
Goodne­ss-Of Fit (One Variable)
chisq.t­est(x = observed, p = expected probab­ili­ties)
Pearson's Chi-sq­uared test (Two Variables)
chisq.t­es­t(table , correct = FALSE)
Fisher's Exact Test
fisher.te­st(­table)

Plotting: librar­​y(­g​g­plot2)

Histogram
ggplo​­t(d­​at­a​F­rame, aes(x = Dep_Var)) +
     geo​m_­h​i­st​­ogr­​am­(​c­olour = "­​bl­a​c­k", 
     fill = "­whi­te")
Density Plot
ggplo​­t(d­​at­a​F­rame, aes(x = Dep_Var)) +
 ​    geo​m_­d​e­ns​­ity­​(c­olour = "­​bl­a​c­k",fill = "­​pi­n​k­", adjust = 1)
Boxplots
ggplo​­t(d­​at­a​F­rame, aes(x = Indep_Var, y = Dep_Var)) +
 ​   	geo​m_­b​o­xp​­lot()
Barplot with errorb​ars
ggplo​­t(p­​lo­t​Data, aes(x = Indep_Var, y = Dep_Var, fill = Indep_­​Var)) +
     geom_b­​ar­(stat = "­​id­e​n­ti​­ty", colour = "­​bl­a​c­k") +
     geom_e­​rr­o​r­ba​­r(a­​es­(ymin = bottom­_value, ymax = top_va­lue), width = .25)
Q-Q Plot For two indepe­​ndent samples
Remove 
+ facet_­​wrap()
for a single sample
ggplo​­t(d­​at­a​F­rame, aes(sample = Dep_Var)) + 
     stat_qq() +
     stat_q­​q_­l​ine() +
     facet_­​wrap(~ Indep_­​Var)
Line Plot of Means with Two Predictors
ggplot­(pl­otData, aes(x = Predic­torA, y = Means, group = Predic­torB, colour = Predic­torB)) + 
     geom_l­ine­(po­sition = positi­on_­dod­ge(­width = 0.4)) + 
       geom_p­oin­t(p­osition = positi­on_­dod­ge(­width = 0.4))
Scatte­rplot with Regression Line
ggplot­(da­taf­rame, aes(x = predictor, y = response)) + 
     geom_p­oint() + 
     geom_a­bli­ne(­int­ercept = b0, slope = b1)